NMDA receptors are critically involved in a variety of CMS functions such as learning, pain amplification, motor pattern generation, and experience-dependent synapse formation and elimination. These receptors are also involved in various neuropathological conditions such as epilepsy, opiate addiction, and neuronal cell death following head/spinal cord injury, sroke, AIDS infection, and possibly the pathophysiology in schizophrenia, Alzheimer's, Parkinson's, and Huntingon's. However, in the absence of selective pharmacological tools, relatively little is known about the role of different NMDA receptor subtypes in these critical cellular processes and disease states. While most of the functional and pharmacological heterogeneity is due to the four genetically-distinct NR2 NMDA receptor subunits, highly-selective antagonists are only available for the NR2B subuit. Over the past few years we have generated a novel series of compounds that are the only NR2C, and NR2D agents known and have already revealed differential roles for NMDA receptor subtypes in synaptic physiology and plasticity. However, it has proven difficult to develop highly-selective agents because the central glutamate binding site is highly conserved for each of the NR2 subunits (as well as for the various AMPA and kainate receptor glutamate binding sites). We have now identified the subunit-specific structural features of the NR2 glutamate binding sites and we propose a two-step approach to developing NR2D and NR2C subtype-selective antagonists. 1) Use a combination of molecular modeling and site-directed mutagenesis of recombinant receptors tested in Xenopus oocytes to define the precise placement of a select group of NMDA antagonists that are large enough to reach the subunit-specific amino acid residues at the edge of the glutamate-binding pocket. 2) Using this structural information, design and test antagonists that are predicted to interact with amino acid residues that are specific to NR2D and other NR2 subunits. Using this approach, we have already designed several antagonists that are predicted by molecular modeling techniques to be highly-selective for NR2D subunit-containing NMDA receptors. [unreadable] [unreadable] [unreadable]